Predictors and outcomes of converted minimally invasive pancreaticoduodenectomy: a propensity score matched analysis

Background: Data-driven patient selection guidelines are not available to optimize outcomes in minimally invasive pancreaticoduodenectomy (MIPD). We aimed to define risk factors associated with conversion from MIPD to open PD and to determine the impact of conversion on post-operative outcomes. Methods: We conducted a retrospective review of MIPD using NSQIP from 2014 to 2015. Propensity score was used to match patients who underwent completed MIPD to converted MIPD. Results: 467 patients were included: 375 (80.3%) MIPD and 92 (19.7%) converted. Converted patients were more often male (64% vs. 52%, p = 0.030), had higher rates of dyspnea (10% vs. 3%, p = 0.009), underwent more vascular (44% vs. 14%, p < 0.001) or multivisceral resection (19% vs. 6%, p = 0.0005), and were more likely attempted laparoscopically compared to robotically (76% vs. 51%, p < 0.001). Robotic approach was independently associated with reduced risk of conversion (OR 0.40, 95% CI 0.23–0.69), while male gender (OR 1.70, 95% CI 1.02–2.84), history of dyspnea (OR 3.85, 95% CI 1.49–9.96), vascular resection (OR 4.32, 95% CI 2.53–7.37), and multivisceral resection (OR 2.18, 95% CI 1.05–4.52) were associated with increased risk. Major complications were more common in converted patients (68% vs. 37%, p < 0.001). Converted patients had increased odds of non-home discharge (OR 3.25, 95% CI 1.06–9.97) and an associated increased length of stay of 3 days (95% CI 0.1–6.7). Conclusion: Patients with a history of dyspnea or tumors requiring vascular or multivisceral resection were at increased risk of conversion, and the robotic platform was associated with a lower rate of conversion. Conversion was independently associated with increased overall complications, increased length of stay, and non-home discharge.